Misunderstanding #16: multiple data corruption and repair misunderstandings
Many of the popular versions are incorrect.
I have heard a lot about what data restoration can do, what cannot be done, what can cause data corruption, and whether the damage can disappear on its own. As a matter of fact, I have already written too many blog posts on such issues. Therefore, this blog can be used as a summary of "rumor Terminator". I hope you will gain some benefits.
First, what can be done for data restoration, and what cannot be done, I have written a blog post about Misconceptions around database repair covering 13 misunderstandings-whether DBCC CHECKDB can be used to fix errors (of course not) whether REPAIR_ALLOW_DATA_LOSS will cause data loss (this name is indeed confusing ).
Secondly, many people complain that the errors displayed during the first running of dbcc checkdb will automatically disappear during the second running. This is a good explanation: the error page detected by dbcc checkdb For The First Time does not belong to the page allocation set, so it cannot be displayed when DBCC is run for the second time. I have a blog post explaining this in detail: Misconceptions around except uptions: can they disappear ?.
Another widely spread rumor is that long-running operations (such as index reconstruction, large data insertion, database or file contraction) can cause page damage. Otherwise, unless SQL Server has bugs (very rare ). Without any T-SQL statements, data errors may occur. I wrote an article A few years ago to explain this in detail: Search Engine Q & A #26: Myths around causing upload uption.
I hope this article will help clarify this concept.